Machine tool monitoring using data analytics and physics-based models

为了保持商业竞争力，Checkfluid不断寻求并实施最佳的制造方法。该项目的重点是如何降低制造成本，同时最大限度地提高产品质量和减少对环境的影响。Checkfluid希望利用他们现有的专业知识和不断增长的在线动态信号分析技术来检测、诊断和预测金属加工操作中的问题（机床磨损，零件生产质量低，机器部件和/或系统退化），并将联合收割机这些专业领域与新的数据分析相结合（用于改进故障检测和诊断）和基于物理的部件和系统建模（用于改进故障和退化预测）。第一个目标是最终确定和验证用于检测和诊断切削刀具性能退化的精确数据驱动建模工具。第二个目标是定义、测试和验证一个切削刀具分析模型，该模型可用于在可变的操作条件下预测切削刀具磨损，并且现有的表示过去刀具磨损率和刀具故障的数据不可用。第三个目标是定义各种策略，将数据驱动的模型和基于物理的模型结合到混合方法中，并测试这些方法，以确认其优于“标准”方法的性能。第四个目标是调查不同的机器（如齿轮箱）的开发方法的应用，以调查这些新技术的通用应用。在本研究工作期间，将对至少6名HQP进行培训。

In order to remain commercially competitive Checkfluid continually searches for and implements optimum manufacturing methods. The focus of this project is on ways to reduce manufacturing costs while maximizing the product quality and minimizing environmental impact. Checkfluid wishes to leverage their existing expertise with the growing knowledge of on-line dynamic signal analysis techniques for detecting, diagnosing, and predicting problems in metal machining operations (machine tool wear, low quality part production, machine component and/or system degradation) and combine these areas of expertise with new data analytics (for improved fault detection and diagnosis) and physics-based component and system modeling (for improved fault and degradation prediction).The proposed project has four main objectives. The first objective is to finalize and verify accurate data-driven modeling tools for use in detecting and diagnosing degradation of cutting tool performance. The second objective is to define, test and verify a cutting tool analytical model that can be used for cutting tool wear prediction under variable operating conditions and where existing data representing past tool wear rates and tool failures is not available. The third objective is to define various strategies for combining the data-driven models and the physics-based models into hybrid methodologies and test these to confirm their advantageous performance over 'standard' methods. The fourth objective is to investigate the application of the developed methods on different machines (such as gearboxes) to investigate the generic application of these new techniques. A total of at least 6 HQP will be trained during this research work.